The conventional method of clock loss detection tests average frequency during a predetermined period, therefore it may not promptly detect input clock loss fault. Further, there may be one cycle sampling error in start and stop point. Further, in order for the method to work, input clock duty cycle should meet Nyquist sampling theory to find an edge transition. Due to the above disadvantage of existing clock loss detection method, it is desirable to have a new device and method of detecting input clock loss.